effect of growth techniques on the structural, optical and ... · for the rf sputtering system, the...
TRANSCRIPT
1
Effect of Growth Techniques on the Structural, Optical and
Electrical Properties of Indium doped TiO2 Thin Films
Faisal S. Al mashary a, b, *, Suelen de Castro c, Arlon Fernando da Silva d, Jorlandio Francisco Felix d, Marcelo
Rizzo Piton c, Helder Vinícius Avanço Galeti c, Ariano De Giovanni Rodrigues c, Yara Galvão Gobato c, Noor Al
Saqri a, Mohamed Henini a, e, Maryam M. Al huwayz a, Abdulrahman M. Albadri f, Ahmed Y. Alyamani f, Hamad
A. Albrathen f, g, Sami A. Alhusaini f, Khalid M. Aljaber f, Ali Z. Alanazi f, Fahad S. Alghamdi f
a School of Physics and Astronomy, University of Nottingham, Nottingham, NG7 2RD, United Kingdom b Department of Physics, College of Science, Qassim University, Buraydah, 14452, Saudi Arabia c Departamento de Física, Universidade Federal de São Carlos, São Carlos 13560-905, Brazil
d Institute of Physics, Universidade de Brasília, Brasília, DF 70910-900, Brazil e UNESCO-UNISA Africa Chair in Nanoscience & Nanotechnology Laboratories. College of Graduate
Studies, University of South Africa (UNISA), Muckleneuk Ridge, PO Box 392, Pretoria, South Africa f National Center for Nanotechnology and Advanced Materials, King Abdulaziz City for Science and Technology,
Riyadh, 11442, Saudi Arabia g Physics and Astronomy Department, ALASeR/KAIN, King Saud University, Riyadh 11451, Saudi Arabia
* Corresponding author. School of Physics and Astronomy, University of Nottingham, Nottingham, NG7 2RD,
United Kingdom
E-mail address: [email protected] (F. S. Al mashary)
Keywords: In-doped TiO2, Pulsed laser deposition, Sputtering, XRD, Photoluminescence,
Deep Level Transient Spectroscopy
Abstract
We have investigated the effect of the growth techniques on the structural, the
electrically and optically active defects in Indium doped TiO2 thin films grown by pulsed laser
deposition (PLD) and sputtering techniques. X-ray diffraction (XRD) and Raman spectroscopy
patterns revealed both rutile and anatase phases for the sputtering samples. On the other hand,
only the anatase phase was observed for the PLD samples. The photoluminescence (PL) spectra
have unveiled several peaks which were explained by defect related optical transitions.
Particularly, the PL bands are fully consistent with anatase/rutile TiO2 phases and the formation
of In2O3 during the preparation of our samples. It was also observed that at -4 V reverse bias,
the PLD samples have lower leakage currents (~1.4 x 10-7 A) as compared to the sputtering
samples (~5.9 x 10-7 A). In addition, the PLD samples exhibited lower ideality factors and
higher barrier heights as compared to those grown by sputtering. Finally, the Deep Level
CORE Metadata, citation and similar papers at core.ac.uk
Provided by Nottingham ePrints
2
Transient Spectroscopy (DLTS) measurements have shown only one defect in the PLD samples
whereas five defects have been detected in the sputtering samples. Therefore, our results
provide strong evidence that the PLD technique is better suited for the growth of In-doped TiO2
thin films.
1. Introduction
Transparent conducting oxides (TCOs) materials have wide applications in terms of
absorbing, transparency of the visible light and electrical conductivity. They are used in
optoelectronic, photovoltaic, photocatalytic and photoelectrochemical water splitting
applications [1]. TiO2 is a TCO material which has been extensively studied for photocatalysis
and photovoltaic applications. This wide interest in TiO2 is due to its efficient photoactivity,
high stability, low cost and non-toxicity [2-4]. The large band gap (3.0-3.4 eV) [5, 6] of TiO2
makes it only sensitive to UV light and therefore it is not efficient for solar cell applications.
Among a lot of approaches that have been used to modify the electrical and optical properties
of TiO2 in order to make it sensitive to visible light, metal ion doping has been proved to be
one of the best methods [7]. Different metals have been used as dopants in TiO2 to change its
electronic and optoelectronic properties [8]. Using indium as a dopant has attracted a lot of
attention as it can modify some properties of TiO2, such as band gap, surface composition and
charge transport.
Oxygen vacancies are well-known defects in TiO2 which can alter the geometric
structure and the chemical properties of the system [9]. In addition to the defects that may be
introduced into the forbidden band gap of TiO2 due to incorporating indium, the growth
methods could also provide defect levels within the band gap. Understanding the deep and
shallow level defects is essential for future devices [10]. Deep level transient spectroscopy
(DLTS) and photoluminescence (PL) are powerful tools to investigate these defects in
3
semiconductors. Particularly, by using DLTS, defects parameters such as the activation energy
of a defect, its capture cross section and its concentration can be determined. In this study, we
investigated the effect of growth techniques, namely pulsed laser deposition (PLD) and
sputtering, on structural, optical and electrical properties of indium doped TiO2 thin films
deposited on silicon substrates.
2. Sample Details
Two series of samples were prepared for this study using two different growth methods.
For the first series, a layer of 380 nm TiO2 thin film, measured using cross-section by scanning
electron microscopy (SEM, HITACHI S4500), was deposited on (100) n-type silicon substrate
using pulsed laser deposition (PLD/MBE 2100) from PVD products. The laser source was a
KrF excimer Laser (λ = 248 nm, pulse width 20 ns, and repetition rate = 10 Hz) operating at
350 mJ to ablate the target with two sets of pressure (under vacuum: 10-6 Torr and with oxygen
background pressure: 5×10−3 Torr). The growth temperature was fixed at 500 ˚C and the
distance from substrate to target was set at 55 mm. All PLD parameters which contribute to the
growth rate were kept constant during the operating time. This series of samples is named PLD
sample. In the second series, RF magnetron sputtering technique was used to deposit a 300 nm
thin film of TiO2 on (100) n-type silicon substrate. This series is labelled as sputtering sample.
For the RF sputtering system, the samples were deposited using a 2-inche TiO2 target at a
temperature of 500 ˚C in pure Argon ambient without addition of oxygen. The chamber was
evacuated to a high vacuum of less than 1×10-8 Torr. The substrate was rotated during the
deposition at a low speed to enhance the thickness uniformity of the films. Substrates were
mounted on 300 mm stainless steel rotating disk where the distance between target and
substrate was 150 mm. High purity argon gas (99.999%) was introduced at a rate of 22 sccm
as an inert gas for plasma. Before deposition, the samples were pre-sputtered in argon plasma
4
for 10 mins to remove any contaminants. The working pressure was set to 10-3 Torr and RF
power was fixed at 150 W. By thermal evaporation, layers of indium (50 nm thick) were
deposited on top of each sample set in order to keep the same doping concentration of indium
and investigate only the effect of the growth techniques on the electrical and optical properties
of In doped TiO2 thin films. In order to incorporate indium into the TiO2 lattice, both series of
samples were annealed under oxygen flow with a temperature ramp rate of 15°C/minute and a
dwelling temperature of 500 °C for 30 minutes. This annealing process has been found as the
best among many annealing processes and procedures.
A third series of control samples, which do not contain In, were grown by sputtering
and PLD techniques using the same procedures as described above. This will allow us to
investigate the effect of In doping. These samples are labelled here as Reference samples
To prepare the samples for electrical characterization, the devices were processed in the
form of circular mesas with a diameter of 900 μm. The back sides of the silicon substrates were
exposed to evaporated Aluminium to create back ohmic contacts. Ti/Au has been deposited by
thermal evaporation on top of the In-doped TiO2 layer to create Schottky contacts. Fig. 1 shows
the growth and fabrication steps.
Fig. 1 Growth and fabrication steps.
5
The photoluminescence (PL) spectra of In doped TiO2 thin films were investigated as
a function of temperature using a laser excitation wavelength of 325 nm. X-ray diffraction
(XRD) patterns were measured in a Bruker D8 Discover Diffractometer, using Cu Kα radiation
(λ = 1.5418 Å). The diffraction pattern was obtained at diffraction angles between 10º and 80º
with geometry 2θ (the glancing angle X-ray diffraction-GAXRD method) and the conventional
θ-2θ at room temperature. The Raman measurements were performed using a Renishaw micro
Raman inVia spectrometer under ambient conditions. A laser with wavelength of 514 nm was
used as an excitation source.
Current-voltage (I-V) measurements of both series of samples have been performed
using a Keithley 236 source measure unit. Capacitance-voltage (C-V) measurements were
investigated at 300 K with 1 MHz frequency using Boonton 7200 capacitance meter. DLTS
measurements were carried out using a He closed-loop cycle cryostat (Janis CCS-450) and
Boonton 7200 capacitance meter.
3. Results and Discussion
3.1 Structural and optical investigation
Fig. 2 shows the XRD pattern of the PLD and sputtering samples. The XRD θ-2θ pattern
of the sputtering reference sample (Fig. S1 in the supplement) exhibited diffraction peaks at
25.4°and 25.5° corresponding to the (011) and (110) lattice planes revealing the anatase and
rutile phases, respectively. Fig. 2(a) shows the XRD pattern of the In doped TiO2 sputtering
sample (In doping obtained by depositing and annealing an indium layer of 50 nm thickness).
The result shows the presence of anatase and rutile phases of TiO2, however, we can also notice
6
an additional signature of indium oxide phase with the main peaks at 22.5º, 30.6º, 35.6º, 47.3º,
51.1º, and 60.9º corresponding to the (211), (222), (400), (431), (440), and (622) lattice planes
of cubic In2O3 (JCPDS File no:06-0416), respectively. Additionally, the precipitation and
formation of a second phase may contribute to the increasing of trap defects, as will be shown
later using the DLTS technique. TiO2 thin film grown by sputtering technique composed of
rutile and anatase phases have been also observed by Huang at al. [11].
Fig. 2 Room temperature XRD pattern of In doped TiO2 thin films prepared by (a) sputtering and (b)
PLD.
A single rutile phase characteristic of TiO2 thin films is obtained for PLD reference sample
(Fig. S2 in the supplement). Fig. 2(b) shows the XRD pattern of PLD samples. The peaks are
in good agreement with the standard XRD patterns of TiO2 found in the literature [11]. It can
be seen that the peak R(110) strongly dominates over the other peaks related to rutile TiO2
structure, indicating the preferential orientation. However, in the In-doped TiO2 PLD sample
the presence of both indium oxide phase and TiO2 pure- rutile phase are observed (see Fig.
2(b)). From the XRD results, it is possible to conclude that for the sputtering reference sample
both rutile and anatase phases are present, while only rutile phase was detected in PLD
reference sample. Moreover, the presence of a layer of indium (50 nm thickness) causes the
formation of In2O3 phases in both the sputtering and PLD samples. However, the
polycrystalline character of the In2O3 phase is more intense in the sputtering samples. This
7
latter behaviour will have a strong influence on the electrical properties of the devices, as will
be shown later.
In order to confirm the XRD results, Raman studies have been performed. Fig. 3(a)
shows the Raman spectra of the In-doped TiO2 samples obtained by sputtering, where both
rutile and anatase phases are observed. The sputtering sample shows an intense peak at Eg=142
cm-1 from anatase phase. The inset of Fig. 3(a) shows a weak peak at B1g=399 cm-1 and Eg=638
cm-1 that are characteristics of anatase phase of TiO2 [12]. A single band at Eg=446 cm-1 from
TiO2 rutile phase is also observed. Finally, a weak vibrational band has been observed for cubic
In2O3 at 197 cm-1 which is assigned to the O–In–O bending modes of cubic In2O3. The most
intense peaks at 520cm-1 are due to the Raman signal from Si(100) substrates. The results from
XRD also show similar behavior. The same Raman bands are also observed in the sputtering
reference sample (Fig. S3 in the supplement).
Fig. 3 Raman spectrum of In-doped TiO2 thin films prepared by (a) sputtering and (b) PLD.
Fig. 3(b) shows the Raman spectra of PLD samples. As can be seen in Fig. 3(b), only two peaks
at Eg=446 cm-1 and Eg=610 cm-1 are observed which are related to rutile phase of TiO2 [13].
These results are in good agreements with XRD results, where only the rutile phase was
8
observed. The Raman spectrum for PLD reference samples (Fig. S4 in the supplement) showed
the same characteristics.
3.2 Current-voltage measurements
I-V measurements as a function of temperature (20-440 K with 20 K intervals) were
performed in both samples in order to determine the diode parameters such as ideality factor
(n), barrier height (ϕb) and series resistance (Rs). Fig. 4 shows the room temperature semi-
logarithmic plot of I-V characteristics for both samples. It can be seen from the figure that the
reverse current of the sputtering sample is higher than that of PLD sample. In particular, the
reverse bias leakage current at – 4 V for PLD and sputtering samples are 1.3 x 10-7 A and 5.9
x 10-7 A, respectively. This may indicate the presence of more defects in sputtering samples
than PLD samples. These defects which act as generation-recombination centres and contribute
to the tunnelling process in reverse bias can be probed by DLTS technique [8]. In addition,
from the linear I-V plot (inset of Fig. 4) a turn-on voltage (Von) of 0.26 V and 0.77 V was
obtained for the sputtering and PLD sample, respectively.
The I-V characteristics of an ideal diode including series resistance (Rs) could be
described by the thermionic emission model [14] as:
𝐼 = 𝐼0[𝑒𝑥𝑝 (𝑞(𝑉−𝐼𝑅𝑠)
𝑛𝐾𝑇) − 1] (1)
where 𝐼0 is the saturation current and is given by:
𝐼0 = 𝐴𝐴∗𝑇2 𝑒𝑥𝑝 (−𝑞𝜙𝑏
𝐾𝑇) (2)
In Eqs. (1) and (2) A is the effective diode area (𝐴 = 6.362 x 10-3 cm2 for both samples), 𝐴∗ is
the effective Richardson’s constant (𝐴∗ = 671 A cm-2 K-2 [15]), k is Boltzmann’s constant, T
is the temperature, q is the elementary charge, n is the ideality factor, ϕb is the barrier height
and Rs is the series resistance. In order to improve the accuracy of calculating the diode
9
parameters (n, ϕb, Rs and 𝐼0), the Werner’s method was used [16]. These parameters for PLD
and sputtering samples at room temperature are shown in Table. 1.
Fig. 4 Semi-logarithmic I-V plots for PLD and sputtering samples. The inset shows the linear I-V plot.
Table. 1 Ideality factor (𝑛), barrier height (ϕb) and series resistance (Rs) at room temperature for
PLD and sputtering samples.
Sample 𝑛 ϕ𝐵 (eV) R𝑆 (kΩ)
PLD 2.01 0.87 330.86
sputtering 2.54 0.77 5.92
The plot of ln (I0/T2) versus 1000/T for PLD sample (Fig. S5 (a) in the supplement) in
the temperature range 100-420 K follows a straight line. This behaviour suggests that the
conduction mechanism could be governed by thermionic emission and the saturation current
can be obtained using Eq. (2) [17]. A similar behaviour was observed in the sputtering sample
in the temperature range 160-420 K (Fig. S5 (b) in the supplement). The Richardson constants
that are calculated from the intercept of the straight line of ln (I0/T2) versus 1000/T plot for
10
PLD sample (A∗= 1.917 x 10-6A/K2 cm2) and sputtering sample (A∗= 4.956 x 10-8 A/K2 cm2)
are much lower than the well-established value (A∗= 671 A/K2 cm2). This deviation from the
Richardson constant value may be due to the spatial inhomogeneous barrier and potential
fluctuations at the interface that consist of low and high barrier areas [18, 19]. Fig. 5 shows
the temperature dependence of the barrier height (ϕ𝐵
) and ideality factor (𝑛) for both samples.
As the temperature increases, the barrier height increases while the ideality factor decreases.
This behaviour is known to be due to nonuniformity of interfacial charges. At low temperature,
carriers are frozen and therefore the current does not follow the thermionic emission
mechanism. At low temperatures, the current passes through the interface states and this result
in higher values of the ideality factor [20]. Note that the carriers, at low temperature, can
overcome the lower barriers and the mechanism of transport will be dominated by the current
flowing through the region with lower barrier height. As temperature increases, carriers gain
sufficient energy to overcome the higher barriers, which results in an increase of the barrier
height with temperature [21].
The barrier height and ideality factor behaviour with temperature have been attributed
to the barrier inhomogeneity. This inhomogeneity of barrier height was also confirmed by the
almost linear relationship between barrier height and ideality factor for both samples (Fig. S6
in the supplement) [22]. It is worth noting that deviation from a linear relationship has been
attributed to uncertainties in the calculation of the ideality factor [23].
11
Fig. 5 In doped TiO2 samples grown by PLD and sputtering (a) barrier height (ϕ𝐵
) versus temperature;
(b) ideality factor (𝑛) versus temperature.
The main trap which contributes to the leakage current can be calculated using I-V
measurements at different temperatures by plotting the reverse current versus inverse
temperature [20]. Fig. 6 shows the plot of reverse current versus 1000/T for -1 V bias voltage
for both samples. One main trap has been found in PLD sample with an activation energy
EPLD= 0.492 ± 0.007 eV. However, two traps have been detected in sputtering sample, namely
ESP1= 0.100 ± 0.005 eV and ESP2= 0.51 ± 0.04 eV. Note that EPLD and ESP2 have similar
activation energies and the origin of this trap may be due to oxygen vacancy [7] or a deep donor
Ti+3 state [10]. ESP1 may be related to oxygen vacancies or interstitial Ti ion [24]. A large
number of defects determined from I-V measurements could be explained a high value of
reverse current [20]. Since the sputtering sample has a larger number of defects than the PLD
sample that could account of the larger value of the reverse current observed in the sputtering
sample. It is worth noting that the numbers of traps found using the I-V measurements are
consistent with those found by using optical technique, as we will show later.
12
2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5
-20
-18
-16
-14
-12
-10
VR= -1V
EP= (0.492 ± 0.007) eV
ES1= (0.1 ± 0.005) eV
ES2= (0.51 ± 0.04) eV
Linear Fit
Linear Fit
Linear Fit
ln(I
)
1000/T (K-1)
Fig. 6 Reverse current versus 1000/T for reverse bias voltage VR = -1 V for In-doped TiO2 samples
prepared by PLD and sputtering
3.3 Capacitance-voltage measurements
The C-V measurements for both samples at room temperature are shown in Fig. 7(a).
It is clear from the figure that the sputtering sample has higher capacitance than the PLD sample
and this could be due to shallow ionised impurity levels. Also, a small shoulder can be seen in
both samples, at -0.6 V and at -1.3 V for PLD and sputtering, respectively. This behaviour
could be due to a poor interface quality and drift in carriers [8]. The slope of the best fit of the
plot 1/C2 versus reverse bias, Fig. 7(b), was used to calculate the free carrier concentrations,
Nd, for both samples at room temperature. For the PLD sample, the doping concentration was
found to be uniform as evidenced by the linear behaviour of the plot of 1/C2 versus applied
bias. The free carrier concentration is about 5.4 x 1013 cm-3. However, as depicted in Figure 7
(b), for the sputtering sample, two lines were fitted to the C-V data, confirming two different
13
doping concentrations of 1.0x1014 cm-3 and ~3.0x1014 cm-3. The free carrier concentration is
higher near the junction interface than in the bulk.
Fig. 7 (a) C-V plot and (b) 1/C2 versus V plot at room temperature of In-doped TiO2 samples prepared
by PLD and sputtering.
14
3.4 Optical properties
Fig. 8 (a) and (b) show typical normalized PL spectra at 12K and 300K for both samples
prepared by sputtering and PLD. It was observed that the PL spectrum of the sputtering sample
consists of only one emission broad band around 1.95 eV. This result is similar to the one
previously reported for In doped TiO2 films prepared by e-beam evaporation [8]. The nature
of this emission band was previously attributed to the In3+ cation, which may produce a
radiative transition between In (5P) and oxygen defect level [8]. However, the observed broad
PL band could also be associated to optical transitions related to anatase and rutile phases. It
can also include an important contribution of In2O3 which can be formed during the thermal
annealing of the samples to incorporate In. On the other hand, the PLD sample has revealed
several peaks. In order to understand the contribution of In doping we have also studied PL
from reference samples prepared by sputtering and PLD. Fig. 8 (c) and (d) show the normalized
PL spectra at 300K for the reference and doped TiO2 samples prepared by both techniques. We
remark that for the reference sample prepared by sputtering the PL spectra shows four emission
peaks (Fig. 8(d)) at 300K. As will be discussed below, the anatase phase of TiO2 usually shows
PL bands around 1.9 and 2.4eV while the rutile phase usually shows PL bands around 1.5 and
1.7eV [25]. Therefore, the PL peaks observed at 300K for the reference sample prepared by
sputtering are fully consistent with the presence of anatase and rutile phases which were also
evidenced by XRD and Raman results.
As mentioned previously several bands were observed for the PLD sample in the visible
and infrared region at 12K and 300K (Fig. 8(a) and 8(b)) which are labelled as P1 (1.93 eV),
P2 (2.52 eV), P3 (2.90 eV), and P4 (3.25 eV) at 12 K. In the following, we present a detailed
discussion about the nature of these emission bands and possible interpretation of PL peaks
observed in our PLD sample.
15
Fig. 8 Normalized PL spectra of In-doped TiO2 thin films grown by PLD and sputtering at (a) 12K
and (b) 300K. Comparison between PL emission at 300K for undoped and In-doped TiO2 grown by (c)
PLD and (d) sputtering.
As mentioned previously the PL spectra of TiO2 material usually depends on the crystal
phase (anatase, rutile or mixed phases) [25]. It was reported [25] that the TiO2 PL spectrum
shows green (around 2.5 eV) and red (around 1.9 eV) emissions for the anatase phase and only
a near-infrared emission (NIR) band around 1.5 eV for the rutile phase. The green PL emission
is usually attributed to self-trapped excitons [26], oxygen vacancies [27] and surface states
[28]. However, a detailed study of these emissions has demonstrated that the green emission
could also be related to radiative recombination of free electrons with trapped holes. These
trapped holes could be localized at oxygen vacancies sites (VO) or at Ti sites adjacent to VO
[25]. The red PL emission is usually related to recombination between trapped electrons and
free holes [25]. On the other hand, the nature of NIR emission in TiO2 is more difficult to
explain [25]. Usually, the rutile samples are less sensitive to prolonged exposure to UV
16
illumination in oxygen environment than anatase samples, which suggests that the NIR PL
could be associated to a possible radiative recombination between midgap trapped electrons
and free holes at valence band. On the other hand, our XRD and Raman spectroscopy results
have revealed both TiO2 and In2O3 phases for the In-doped TiO2 samples. The formation of
In2O3 after the thermal annealing process which was used to incorporate In in the TiO2 lattice
could also contribute to the PL bands observed in the In-doped TiO2 samples. It is well known
that In2O3 oxide is also a wide band gap (~3.6 eV) semiconductor [29]. It was previously
reported that the PL spectra displayed a near band edge (NBE) emission around 3.23 eV and a
defect-related emission at ~1.98 eV [30]. The PL peak in the visible region ( defect-related
emission) was previously associated to the transition from the oxygen vacancy level to the
valence band[31]. In addition, this red emission was also previously associated with deep level
states (1.94 eV) which can be present in In2O3 [32].
As mentioned above our In-doped TiO2 PLD samples revealed four emission bands
(labelled P1, P2, P3 and P4). It was found that as the temperature increases, a gradual reduction
of P1 emission band was observed as expected. However, additional weak emissions bands
located around 1.5eV and 1.6eV were also detected. The PL peak energies of most of these
bands are smaller than the energy gap of TiO2 and In2O3. Therefore, all these emission bands
were attributed to the localized levels in the bandgap and could be due to intrinsic defects in
both TiO2 and In2O3. As mentioned above, the observed defect emission peaks for TiO2 and
In2O3 usually have emissions in the same energy range. Therefore, it is difficult to separate the
contribution of defect-related emissions from the In2O3 and TiO2. In order to investigate which
peak could be associated to the TiO2, we have also studied a reference sample, i.e. undoped
TiO2 sample. Fig. 8 (c) shows the 300 K PL spectra of undoped and In-doped TiO2 50nm
samples prepared by PLD.
17
As can be seen in Fig. 8(c), the PL peaks of the undoped TiO2 thin film have a smaller
intensity in the range of about 1.8-2.9 eV and around 1.35 eV as compared to those in In-doped
TiO2 samples. The observed bands of 1.8-2.9 eV and ~1.35 eV are assigned to the anatase and
rutile phases (mixed phase), respectively [25]. The visible emission was associated to a
combination of two bands previously reported in the literature for the anatase phase, namely
the green band (~2.5 eV) and the red band (~1.9 eV) [25]. However, the anatase phase was not
evidence for XRD and Raman results. This band could be associated to small contribution
anatase phase (mixed phase sample) that cannot be evidenced by XRD and Raman results but
it is important for optical emission. On the other hand, emissions around 1.5 and 1.7 eV have
been observed only for the rutile phase and associated with the radiative recombination
between midgap trapped electrons and free holes at the valence band [25] and Ti3+ ions [33].
We attribute the broad band around 1.2-1.6eV to the rutile phase. On the other hand, additional
PL peaks (P1, P2, P3 and P4) were observed in PLD sample which were associated to the
formation of In2O3. Particularly, the P4 band that observed around 3.25 eV at 12 K was
associated to the band edge emission of In2O3 [32]. The other PL peaks (P1, P2 and P3) have
important contribution of defect emission due to In2O3.
18
We have also investigated the temperature dependence of the emission bands for both
PLD and sputtering samples. Figs. 9 (a) and (b) show a typical temperature dependence of PL
spectra for these samples. It was found that the emission intensity of the peak at 1.93 eV (P1)
decreases considerably with the increase of temperature for both sputtering and PLD samples
(please see Fig. 9(a)) as expected. However, the intensities of the other emissions of the PLD
sample increase when the temperature increases (in the range of 100 to 300K). We believe that
this effect could be associated to charge transfer, or traps due to the presence of defects or some
disorder in TiO2 and/or In2O3. Traps could localize carriers non-radiatively, and the trapped
carriers are thermally activated and contribute to the optically active states. We have also
estimated the activation energy using a double-channel activation energies function [34, 35].
𝐼(𝑇) =𝐼0
1+𝛾1exp (−𝐸𝑎1/(𝑘𝐵𝑇))+𝛾2exp (−𝐸𝑎2/(𝑘𝐵𝑇)) (3)
where 𝐼𝑜 is the PL intensity at T = 0 K, 𝛾1,2are two constants related to the ratio between
radiative and non-radiative recombination processes, 𝐸𝑎1,2 are the activation energies
corresponding to the non-radiative recombination process, and 𝑘𝐵 is the Boltzmann constant.
19
The inset in Fig. 9(a) and (b) shows the temperature (1/kBT) dependence of the integrated PL
intensity (I) (Arrhenius plot) in the temperature range T = 12–300 K for PLD sample (P1=1.93
eV) and sputtering sample (peak at 1.95eV). Clearly, the emission intensity decreases with
increasing temperature following an Arrhenius-like function. Equation (3) has been used to fit
the PL intensity in the whole measured temperature range (dash line). Activation energies of
𝐸𝑆𝑃𝑎1 =82.0 meV and 𝐸𝑆𝑃𝑎2 =12.0 meV, and 𝐸𝑃𝐿𝐷𝑎1 =22.3 meV were obtained for sputtering
and PLD thin films, respectively. Al Saqri et al. [8] reported a defect in In-doped TiO2 thin
films with an activation energy of 62 meV. They attributed this defect to the intrinsic Ti atoms
and ionization energies of oxygen vacancies. In our case, the calculated activation energy could
also be related to the similar nature which has already been evidenced from emission processes.
Fig. 9 PL spectra of (a) sputtering sample and (b) PLD sample at different temperatures in the range
12-300K. The insert shows the Arrhenius plot of total intensity for both samples.
20
In conclusion, we have observed well resolved PL bands in the PLD sample and a broad
PL band in the sputtering sample. These results provide strong evidence that PLD samples are
of higher quality than sputtering samples. However, important contribution defect emission
from the formation of In2O3 was observed in the PL from PLD samples. These results are also
consistent with the I-V results for PLD samples which have shown better performance than the
sputtering samples. The optical results are also consistent with the DLTS results which will be
presented below.
3.5 DLTS and Laplace DLTS measurements
In order to investigate the effect of the growth techniques on the electrically active
defects, DLTS technique has been used [36]. The experiment was carried out with a reverse
bias, VR = -1V, filling pulse height, VP = 0V, filling pulse time, tP = 1 msec and rate windows
200 s-1 for both samples. Fig. 10 shows the DLTS signals versus temperature for both samples
over the scanned temperature range 10K-450 K. Broad peaks due to electron traps have been
detected in both samples as shown in Fig. 10. Laplace DLTS measurements [37] have been
carried out on both samples in order to resolve the broader peaks. In the Laplace DLTS spectra
(Fig. S7 in the supplement) only one electron trap (EP) and five electron traps (Es1, Es2, Es3, Es4,
and Es5) were detected in the PLD and sputtering samples, respectively. The activation energies
of these traps were calculated from the Arrhenius plots as shown in Fig. 11. The trap
parameters, such as activation energies, trap concentrations and capture cross sections are
summarised in Table.2.
As seen in Table. 2, traps Ep and Es4 have the same energies so they may have the same
origin which could be assigned to oxygen vacancy as reported by E. Wang et al. [7] in In-doped
TiO2 doped using the sol-gel method. However, B. Morgan and G. Watson have investigated
the formation of native defects in anatase TiO2 using the density functional theory [10] and
21
assigned a trap at 0.5 eV to Ti+3 state. Ep and Es4 could have the same origin. Es1 is a shallow
level trap and Es2 may be related to oxygen vacancy or interstitial Ti ion [24]. The Es2 trap was
also observed using PL measurement, labeled as 𝐸𝑆𝑃𝑎1. Es3 and Es5 may be assigned to a Ti
donor level [38].
Fig. 10 DLTS signal for PLD and sputtering samples.
Fig. 11 Arrhenius plots from Laplace DLTS with reverse biases, VR= -1 V, filling pulse height, VP =
0V, and filling pulse time, tP = 1 msec, for (a) PLD sample and (b) sputtering sample
22
Table. 2 Traps parameters for PLD and sputtering samples at VR= -1V, VP= 0V and tP= 1 msec.
DLTS measurements using different reverse biases are usually performed to control the
depletion region width (moving it away or close to the interface) where the process of trapping
and de-trapping of carriers happen. DLTS measurements have been carried out for PLD sample
at different reverse biases VR = -1, -3, and -4V, as shown in Fig. 12. The filling pulse height,
VP = 0V and the filling pulse time, tP = 1 msec, were kept the same for all measurements. As
we change the reverse bias VR, only one trap has been detected for each reverse bias. Table 3
shows the trap parameters for VR= -1, -3 and -4 V. Arrhenius plots for VR= -3 and -4 V are
displayed in Fig. 13. The trap found at VR= -3 V has the same activation energy as that at VR=
-4 V whose origin could be related to the Ti donor [38]. Note that the amplitude of the DLTS
signal is decreasing as the reverse bias increases (away from the interface). This means that the
concentration of the trap is decreasing as moving away from the interface.
Sample
Trap
Activation
Energy
(eV)
Capture Cross-
Section (σ∞)
(cm2)
Trap Concentration
(cm−3)
PLD Ep 0.469 ± 0.009 3.998 x 10-16 3.749 x 1013
sputtering
Es1 0.0027 ± 0.0001 1.958 x 10-22 5.027 x 1012
Es2 0.096 ± 0.005 6.629 x 10-21 3.939 x 1013
Es3 0.21 ± 0.02 4.013 x 10-17 9.187 x 1013
Es4 0.415 ± 0.001 5.919 x 10-13 1.585 x 1014
Es5 0.324 ± 0.003 8.975 x 10-18 1.341 x 1014
23
Table. 3 Traps parameters for PLD sample at VR= (-1, -3 and -4) V, VP= 0V and tP= 1 msec.
Reverse bias
(V)
Trap
Activation Energy
(eV)
Capture Cross-
Section (σ∞)
(cm2)
Trap
Concentration
(cm−3)
-1 Ep 0.469 ± 0.009 3.998 x 10-16 3.749 x 1013
-3 Ep1 0.335 ± 0.003 1.171 x 10-17 2.854 x 1013
-4 Ep1 0.318 ± 0.002 7.779 x 10-18 2.101 x 1013
Fig. 12 DLTS signal for PLD sample at VR = -1, -3, and -4V
24
Fig. 13 Arrhenius plots for PLD sample obtained from Laplace DLTS at (a) VR= -3 V and (b) VR= -4
V.
4. Conclusion
The effect of the growth technique on the structural, electrically and optically active
defects in In-doped TiO2 has been investigated using two TiO2 sample sets prepared by PLD
and sputtering techniques. The samples grown by sputtering technique have shown more
defects than those grown by PLD. This could be related to the coexistence of both anatase and
rutile phases present in the thin film deposited by sputtering. As a consequence, the I-V
characteristics show lower reverse current for PLD sample as compared with the sputtering
sample. PLD sample has a better value of ideality factor than sputtering sample. One defect
has been found in PLD sample set where five defects have been detected in the sputtering
sample set. Due to the lower leakage currents and less number of defects in PLD sample, it is
concluded that the PLD technique is better suited for the growth of In-doped TiO2.
Acknowledgements
The author would like to thank Qassim University, Saudi Arabia, for funding his PhD
studies. The authors also acknowledge the financial supports from the Brazilian agencies
(Fundação de Amparo a Pesquisa do Estado de São Paulo (FAPESP grants 2016/10668-7 and
25
2014/50513-7) and Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)
(grant 3181/2014-1) and (FAPDF)17077.78.35088.2504/2017.
26
References
1. Stadler, A., Transparent conducting oxides—An up-to-date overview. Materials,
2012. 5(4): p. 661-683.
2. Hashimoto, K., H. Irie, and A. Fujishima, TiO2 photocatalysis: a historical overview
and future prospects. Japanese journal of applied physics, 2005. 44(12R): p. 8269.
3. Sarkar, M.B., et al., Enlarged broad band photodetection using Indium doped TiO2
alloy thin film. Journal of Alloys and Compounds, 2014. 615: p. 440-445.
4. Chinnamuthu, P., et al., Band gap enhancement of glancing angle deposited TiO2
nanowire array. Journal of Applied Physics, 2012. 112(5): p. 054315.
5. Amtout, A. and R. Leonelli, Optical properties of rutile near its fundamental band
gap. Physical Review B, 1995. 51(11): p. 6842.
6. Tang, H., et al., Urbach tail of anatase TiO 2. Physical Review B, 1995. 52(11): p.
7771.
7. Wang, E., W. Yang, and Y. Cao, Unique surface chemical species on indium doped
TiO2 and their effect on the visible light photocatalytic activity. The Journal of
Physical Chemistry C, 2009. 113(49): p. 20912-20917.
8. Al Saqri, N.A., et al., Investigation of defects in indium doped TiO 2 thin films using
electrical and optical techniques. Journal of Alloys and Compounds, 2017. 698: p.
883-891.
9. Nowotny, J., T. Bak, and M.A. Alim, Dual mechanism of indium incorporation into
TiO2 (Rutile). The Journal of Physical Chemistry C, 2014. 119(2): p. 1146-1154.
10. Morgan, B.J. and G.W. Watson, Polaronic trapping of electrons and holes by native
defects in anatase TiO 2. Physical Review B, 2009. 80(23): p. 233102.
11. Huang, H.-H., et al., Preparation of rutile and anatase phases titanium oxide film
by RF sputtering. Journal of nanoscience and nanotechnology, 2008. 8(5): p. 2659-
2664.
12. Mikami, M., et al., Lattice dynamics and dielectric properties of TiO 2 anatase: a
first-principles study. Physical Review B, 2002. 66(15): p. 155213.
13. Lukačević, I., et al., Lattice dynamics and Raman spectrum of rutile TiO2: The role
of soft phonon modes in pressure induced phase transition. Materials Chemistry
and Physics, 2012. 137(1): p. 282-289.
14. Sze, S.M. and M.K. Lee, Semiconductor devices : physics and technology. 2013,
Singapore: Wiley.
15. Pillai, P., et al., Extraction of Schottky barrier at the F-doped SnO2/TiO2 interface
in Dye Sensitized solar cells. Journal of Renewable and Sustainable Energy, 2014.
6(1): p. 013142.
16. Werner, J.H., Schottky barrier and pn-junctionI/V plots—Small signal evaluation.
Applied Physics A: Materials Science & Processing, 1988. 47(3): p. 291-300.
17. Van, C.N. and K. Potje-Kamloth, Electrical and NO x gas sensing properties of
metallophthalocyanine-doped polypyrrole/silicon heterojunctions. Thin Solid Films,
2001. 392(1): p. 113-121.
18. Jameel, D., et al., Electrical performance of conducting polymer (SPAN) grown on
GaAs with different substrate orientations. Applied Surface Science, 2016. 387: p.
228-236.
19. Aydoğan, Ş., M. Sağlam, and A. Türüt, On the barrier inhomogeneities of
polyaniline/p-Si/Al structure at low temperature. Applied surface science, 2005.
250(1): p. 43-49.
20. Aziz, M., et al., Electrical Behavior of MBE Grown Interfacial Misfit GaSb/GaAs
Heterostructures With and Without Te-Doped Interfaces. IEEE Transactions on
Electron Devices, 2015. 62(12): p. 3980-3986.
21. Tripathi, S. and M. Sharma, Analysis of the forward and reverse bias IV and CV
characteristics on Al/PVA: n-PbSe polymer nanocomposites Schottky diode. Journal
of Applied Physics, 2012. 111(7): p. 074513.
27
22. Korucu, D., A. Turut, and H. Efeoglu, Temperature dependent I–V characteristics
of an Au/n-GaAs Schottky diode analyzed using Tung’s model. Physica B:
Condensed Matter, 2013. 414: p. 35-41.
23. Al Saqri, N., et al., Investigation of the effects of gamma radiation on the electrical
properties of dilute GaAs 1− x N x layers grown by Molecular Beam Epitaxy. Current
Applied Physics, 2015. 15(10): p. 1230-1237.
24. Miyagi, T., et al., Photocatalytic property and deep levels of Nb-doped anatase TiO2
film grown by metalorganic chemical vapor depostion. Japanese journal of applied
physics, 2004. 43(2R): p. 775.
25. Pallotti, D.K., et al., Photoluminescence Mechanisms in Anatase and Rutile TiO2.
The Journal of Physical Chemistry C, 2017. 121(16): p. 9011-9021.
26. Skowronski, L., R. Szczesny, and K. Zdunek, Optical and microstructural
characterization of amorphous-like Al 2 O 3, SnO 2 and TiO 2 thin layers deposited
using a pulse gas injection magnetron sputtering technique. Thin Solid Films, 2017.
632: p. 112-118.
27. Iijima, K., et al., Influence of oxygen vacancies on optical properties of anatase TiO
2 thin films. Journal of Luminescence, 2008. 128(5): p. 911-913.
28. Yang, Y., et al., Visible and near-infrared electroluminescence from TiO2/p+-Si
heterostructured device. AIP Advances, 2014. 4(4): p. 047109.
29. Kim, H.S., et al., High performance solution-processed indium oxide thin-film
transistors. Journal of the American Chemical Society, 2008. 130(38): p. 12580-
12581.
30. Chan, C.-H., et al., Optical characterization of structural quality in the formation of
In2O3 thin-film nanostructures. The Journal of Physical Chemistry C, 2016.
120(38): p. 21983-21989.
31. Meléndez, J.J. and M. Wierzbowska, In2O3 doped with hydrogen: electronic
structure and optical properties from the pseudopotential Self-Interaction
Corrected Density Functional Theory and the Random Phase Approximation. The
Journal of Physical Chemistry C, 2016. 120(7): p. 4007-4015.
32. Ghosh, A., et al., Detailed investigation of defect states in Erbium doped In2O3 thin
films. Materials Research Bulletin, 2018. 99: p. 211-218.
33. Johannsen, S.R., et al., Influence of TiO 2 host crystallinity on Er 3+ light emission.
Optical Materials Express, 2016. 6(5): p. 1664-1678.
34. Krustok, J., H. Collan, and K. Hjelt, Does the low-temperature Arrhenius plot of the
photoluminescence intensity in CdTe point towards an erroneous activation energy?
Journal of applied physics, 1997. 81(3): p. 1442-1445.
35. Wang, X., et al., Trap states and carrier dynamics of TiO 2 studied by
photoluminescence spectroscopy under weak excitation condition. Physical
Chemistry Chemical Physics, 2010. 12(26): p. 7083-7090.
36. Lang, D., Deep‐level transient spectroscopy: A new method to characterize traps in
semiconductors. Journal of applied physics, 1974. 45(7): p. 3023-3032.
37. Dobaczewski, L., A. Peaker, and K. Bonde Nielsen, Laplace-transform deep-level
spectroscopy: The technique and its applications to the study of point defects in
semiconductors. Journal of Applied Physics, 2004. 96(9): p. 4689-4728.
38. Salama, A. and L. Cheng, The Effects of Titanium Impurities in N+/P Silicon Solar
Cells. Journal of The Electrochemical Society, 1980. 127(5): p. 1164-1167.